The effects of specific and common variance on test–retest and self-other correlations: Another perspective on personality nuances

Abstract

To explain why the test–retest correlation of a personality measure can exceed its internal reliability, McCrae (2015) proposed that specific variance provides additional reliable information beyond common variance, which he termed personality nuances. Although the definition of nuances as specific variance was later replaced by narrow personality traits measured using a few items, this study laid the foundation for the personality nuances research field. After analyzing the shortcomings of the method of residuals and returning to statistically sound definitions of common and specific variance, we found that the common variance in the facet subscales was nearly double that of the specific variance. At the domain level, the proportion of common variance’s impact was up to nine times greater than that of specific variance, as indicated by their Cronbach alphas. Nevertheless, the specific variance had a much more significant impact on the strength of test–retest correlations and self-other agreement. Therefore, specific variance can be seen as a fundamental unit of personality and its assessment, despite test constructors’ efforts to maximize the amount of common variance in their scales.